BULLETIN

BULLETIN OFOF THE POLISH ACADEMY

THE POLISH ACADEMY OF OF SCIENCES
SCIENCES THERMODYNAMICS, MECHANICAL
TECHNICAL SCIENCES, Vol. 68, No. 3,
TECHNICAL SCIENCES, Vol. 68, No. 3, 2020 2020 AND AERONAUTICAL ENGINEERING
DOI:
DOI: 10.24425/bpasts.2020.133382
10.24425/bpasts.2020.XXXXXX

Vibration analysis of ball bearing considering waviness

under high
under high speed
speed and
and an
an axial
axial load
load
P.P. HOU 1, L.Q. WANG 1, 2*, and Q.Y. PENG 3
1 P.P. HOU1 , L.Q. WANG1, 2∗ and Q.Y. PENG3
MIIT Key Laboratory of Aerospace Bearing Technology and Equipment, Harbin Institute of Technology, Harbin, China.
1MIIT Key 2Laboratory of Aerospace
State Key Laboratory ofBearing Technology
Robotics andHarbin
and System, Equipment, Harbin
Institute Institute of Harbin,
of Technology, Technology, Harbin, China
China.
2 State Key
3 Laboratory of Robotics and System, Harbin Institute of Technology, Harbin, China
GUANGZHOU HAO ZHI INDUSTRIAL CO., LTD., Guangzhou,
3 GUANGZHOU HAO ZHI INDUSTRIAL CO., LTD., Guangzhou, China
China

Abstract. Based on the rolling bearing vibration measurement principle in ISO standard, a nonlinear dynamic model of ball bearing is built and
Abstract. Based on the rolling bearing vibration measurement principle in ISO standard, a nonlinear dynamic model of ball bearing is built and
motion equations of the inner ring, outer ring, and rolling elements are derived by using Lagrange’s equation. The ball bearing model includes
motion equations of the inner ring, outer ring, and rolling elements are derived by using Lagrange’s equation. The ball bearing model includes
the influence of waviness, rotational speed, external load, arbor supporting stiffness and arbor eccentricity. Ball bearing high-speed vibration
the influence of waviness, rotational speed, external load, arbor supporting stiffness and arbor eccentricity. Ball bearing high-speed vibration
tests are performed and used to verify the theoretical results. Simulated results showed that specific waviness orders produced the principal
tests are performed and used to verify the theoretical results. Simulated results showed that specific waviness orders produced the principal
frequencies that were proportional to rotational speed. Rotational speed mainly affected the value of the natural frequency of the bearing system,
frequencies that were proportional to rotational speed. Rotational speed mainly affected the value of the natural frequency of the bearing system,
and RMS (Root Mean Square) of the full band had a great fluctuation with the increase of rotational speed. In the experiment, spectrum and
and RMS (Root Mean Square) of the full band had a great fluctuation with the increase of rotational speed. In the experiment, spectrum and
RMS of 2 fs-30 kHz ( fs: the rotational frequency of inner ring/arbor) under high speed could include not only the influence of rotational speed
RMS of 2 fs -30 kHz ( fs : the rotational frequency of inner ring/arbor) under high speed could include not only the influence of rotational speed
but also principal frequencies produced by waviness, which could cover the part of requirements of the standard bearing vibration measurement.
but also principal frequencies produced by waviness, which could cover the part of requirements of the standard bearing vibration measurement.
Key words: rotational
Key words: rotational speed,
speed, ball
ball bearing,
bearing, vibration
vibrationcharacteristics,
characteristics,frequency
frequencyband,
band,RMS.
RMS.

1. Introduction a)

Bearing vibration measurement is an indispensable process of

bearings quality inspection after their assembly. The rotational
speed is one of the most crucial parameters for bearing vi-
bration measurement. At present, rolling bearings-measuring
methods for vibration have been standardized, for instance, BS
ISO 15242, DIN ISO 15242, and GB/T 24610. Bearings vi-
bration is usually measured with the outer ring being station-
ary and with inner ring turning at a constant rotational speed.
The rolling bearing vibration measurement principle is illus-
trated in Fig. 1. The test ball bearing is usually subjected to an
axial load and its default rotational speed is usually 30 rev/s.
Default analyzed frequency range is 50–10 000 Hz, which is b)
further divided into 50–300 Hz (low frequency band), 300–
1800 Hz (medium frequency band), and 1800–10 000 Hz (high
frequency band). However, the operating speed of modern high
speed rotating machinery (for example, high speed motorized
spindle, aero-engine) is much higher than 30 rev/s. To meet the
speed requirements of modern bearing vibration measurement,
bearing high-speed vibration test rigs are built. J.A. Wens-
ing [1] replaced hydrodynamic bearings of the spindle with an
aerospace bearing, and the standard test speed could be adjusted
within 141.7 rev/s. Adamczak et al. [2] built the Anderometer
STPPD bearing vibration test equipment, whose speed had im-
proved to 60 rev/s. This equipment met the ISO design specifi- Fig. 1. Rolling bearing vibration measurement principle: a) contact
cations. measurement, b) noncontact measurement

∗ e-mail: lqwanghit@163.com
*e-mail: lqwanghit@163.com Bearing vibration is the result of the displacements induced
Manuscript submitted
Manuscript submitted2019-08-21, revisedinitially
20XX-XX-XX, 2020-01-11, initiallyfor
accepted accepted
publication by error motion, which is a consequence of geometric imper-
for publication 2020-01-17,
20XX-XX-XX, published published in June2020.
in ZZZZZZZZ 2020 fections of the diverse internal bearing surfaces. Waviness is an

Bull. Pol. Ac.:

Bull. Pol. Ac.:Tech.
Tech.68(3)
68(3)2020
2020 517
1
 P.P.
P.P. Hou,
Hou, L.Q.
L.Q. Wang,
Wang, and Q.Y. Peng
important part of geometric imperfections and has a great in-
fluence on the characteristics of the bearing vibration response.
There have been many studies that theoretically analyzed the
relationship between waviness and bearing-rotor system vibra-
tion response spectrum. Reference [3, 4] built the relationship
between the waviness order and principal frequencies by solv-
ing the static equations of bearing’s inner ring, for example,
principal frequencies n × Nb × fc (n: integer, Nb : number of
rolling element, fc : cage rotating frequency) were caused by
outer raceway waviness order n × Nb ± 1 and principal frequen-
cies n × Nb × ( fs − fc ) ± fs were generated by inner raceway
waviness order n × Nb ± 1 in the radial direction [4]. In [5], a
rigid shaft supported by a pair of angular contact ball bearings
was modeled as a three degree-of-freedom (3DOF) spring-mass
system in the radial x and y directions and axial z-direction.
The displacement response spectra characteristics due to the
waviness of the outer race, inner race, and rolling element
were analyzed. N. Tandon and A. Choudhury [6] presented a
3DOF dynamic model of the rotor-bearing system including the
movements of the inner ring/shaft, outer ring/bearing house and
cantilever portion/load in the radial x-direction. Simulated re-
sults showed that special frequencies caused by raceways wavi-
ness were related to outer/inner race defect frequency, and that
the amplitudes of the spectral components owing to inner race
waviness were much smaller than these of the outer race wavi-
ness. G.H. Jang and S.W. Jeong [7] considered the 2-angular
motions and the 3-translational motions of the rotor and built
a 5DOF nonlinear dynamic model for a rigid rotor supported
by two or more ball bearings. They indicated that the waviness
between the bearing components and the waviness between
the supporting bearings could interact, and that the centrifugal
force and gyroscopic moment of the ball affected the principle
frequencies [8, 9]. Babu et al. [10] added the angular motion
around the axial z-direction and a 6DOF model of a rigid rotor
was built. The model predicted that the influence of inner race
waviness on the vibration was larger as compared to that caused
by the outer race and rolling element. Yaobin Zhuo et al. [11]
simplified the shaft-bearing system to a 3DOF system, whose
generalized displacements were three translational motions in
the radial x-and
x and y-directions,
y directions, and axial z-direction. Simulated
results revealed that the inner/outer raceway waviness would
excite resonance when the order of waviness was equal or mul-
tiple to Nb . S.P. Harsha and C. Nataraj [12] used Lagrange’s
equation to obtain (Nb + 2) second order, non-linear differen-
tial equations of Nb rolling elements and inner ring in the radial
plane. They analyzed the influence of outer raceway waviness
and the number of rolling elements on the dynamic behavior of
a rotor-bearing system. Combination Lagrange’s equation and
component mode set (CMS) method, J.A. Wensing [1] consid-
ered that the outer ring was flexible. The effect of the wavi-
ness on the response of the outer ring was discussed without
considering the motions of the inner ring. The statistical vari-
ations of vibration generated by the inner ring and ball wavi-
ness were much larger than for vibration generated by outer
ring waviness.
Waviness also has an influence on the bearing vibration level.
In [13], simulated results at the shaft speed of 15 rev/s showed
518
2 Bull. Pol.
and Q.Y. Peng
that the vibration and acoustic response of bearing house would
increase with the growth of waviness size. Adamczak and
Zmarzły [14] carried out 6304-type bearing vibration on the
Anderometer and studied the influence of raceway waviness
on bearing vibration level. The experimental results at 30 rev/s
indicated that an increase in the raceway waviness caused an
increase in the vibration level. For medium bandwidth (300–
1800 Hz), the influence was most prominent. Hiroyuki Ohta
and Shiya Satake [15, 16] formed the vibration test of hybrid
ceramic ball bearings, steel ball bearing and all-ceramic ball
bearing on the Anderometer. The overall value of velocity was
influenced by rotational speed, and it tended to increase with ro-
tational velocities, whose range was from 8.3 to 30 rev/s. How-
ever, some studies had indicated that rotational speed was non-
linear with the dynamic performance of the bearing system [17–
19]. Reference [18] showed that the curve of peak-to-peak of
overall response-speed had a very rough appearance. Peak-to-
peak of 58.3 rev/s was more than that of 166.7 rev/s.
In addition to the bearing models described above for the
study of the effects of waviness on bearing vibrations, a se-
ries of bearing models have also been proposed in the study
of bearing vibration responses with the localized defect. Sidra
Khana et al. [20] built the 2 second-order, nonlinear differen-
tial equations for the inner ring/shaft and outer ring/housing
in the radial x-direction. Sadok Sassi et al. [21] proposed a
3DOF system for the inner ring, ball and outer ring in the
radial y-direction. N.S. Feng et al. [22], and Dick Petersen
and Carl Howard [23] considered the movements of the inner
and outer rings in the radial x- and y-directions and built the
4DOF nonlinear dynamic model for the bearing. N. Sawalhi
and R.B. Randall [24] added a high frequency response of
the bearing in the radial y-direction into the model of [22].
A 5DOF bearing-pedestal model was depicted. This bearing
model was also used in [25]. Dick Petersen et al. [26] con-
tinued to increase the radial x-direction motion of a high fre-
quency response into the model developed by [24]. A 6DOF
bearing-pedestal model was described. Liu et al. [27] pre-
sented a 12DOF model of a rigid rotor-angular contact ball
bearings system. This model included the three translational
and three rotational displacements of the inner race/rotor, three
translational displacements of the outer race of the right bear-
ing, and three translational displacements of the outer race of
the left bearing in the radial x-and
x and y-directions,
y directions, and axial z-
direction.
In application, high speed has become the trend for rolling
bearing vibration measurement, which will accompany the
problem that the divided frequency bands at 30 rev/s may no
longer be suitable for bearing high speed vibration measure-
ment analysis. Therefore, it is necessary to study theoretically
and experimentally the change of vibration characteristics of
the bearing with rotational speed increases, which benefits to
confirm the analysis frequency bandwidth of the evaluation
method of rolling bearing high speed vibration measurement.
However, the limitation of the above studies is to pay more at-
tention to the influence of the special order of waviness of one
of the bearing components on the bearing vibration response.
The experimental and theoretical analyses for rolling bearing
Bull. Pol.Ac.:
Ac.:Tech.
Tech.68(3)
68(3) 2020
2020
 Vibration
Vibration analysis
analysis of of ball
ball bearingconsidering
bearing considering waviness
vibration measurements are rarely performed at the same time,
and for the individual bearing vibration measurement test, the
test speed range can rarely cover the range from low speed to
high speed.
In order to include the influence of rotational speed and sat-
isfy the principle of rolling bearing vibration measurement, the
ball bearing model and the derivation of the motion equations
of the bearing components refer to [1, 12, 18, 28–30]. In this
paper, 2Nb + 5 differential equations of ball bearing are estab-
lished by using Lagrange’s equation. Waviness, operation pa-
rameters (rotational speed and load) and system parameters (ar-
bor supporting stiffness and eccentricity) are considered. Then,
considering bearing components waviness, the influence of ro-
tational speed on response spectra and RMS of ball bearing vi-
bration are analyzed theoretically, which are verified by bear-
ing vibration experiments at low, medium and high speed. At
last, the analysis frequency bandwidth at high speed is dis-
cussed.
2. Ball bearing vibration model considering
rotational speed
The ball bearing is considered a spring-mass system as shown
in Fig. 2. Bearing displacement error should be mainly caused
by the inherent characteristics of bearing components, not the
deformation of them. An appropriate axial load is sufficient to
prevent slipping of the rolling elements relative to raceways and
does not cause deformation of bearing components due to ex-
cessive load. Therefore, the outer ring is not bent, which is dif-
ferent from [1, 29]. The outer ring can move freely in the axial
z-direction, which is different from [12, 18, 28, 30]. Based on
the assumptions in [1, 12, 18, 28–30], the features of the ball
bearing model are as follows:
(1) Ball bearing has a stable rotation. The angular velocity
of the arbor (ωi ) around the z-axial is considered a constant
value. The inner ring is assumed to be rigidly mounted on the
arbor, so the inner ring and arbor have the same displacement
Fig. 2. Spring-mass model of ball bearing
Bull. Pol. Ac.:
Bull. Pol. Ac.:Tech.
Tech.68(3)
68(3)2020
2020
wavinessunder
underhigh speed
high andand
speed an axial loadload
an axial
and angular velocity, ωi . The linear supporting stiffness in the
radial x and y directions is Ksx and Ksy , respectively.
(2) Except for local deformations between the outer-/inner-
raceways and the rolling elements, the outer ring, inner ring,
and rolling elements are rigid. The Hertzian theory of elas-
ticity is used to calculate the local deformations. Rolling el-
ements act as nonlinear contact springs as shown in Fig. 2.
K ob
j is the stiffness of the nonlinear spring connecting the jth
rolling element and outer raceway, and K ib j is the stiffness of
the nonlinear spring connecting the jth rolling element and in-
ner raceway.
(3) The angle velocity of the cage (ωc ) is also assumed to be
constant. The rolling elements are evenly distributed and guided
by the cage. There is no slipping or sliding between the bearing
components. The angular velocity of each rolling element about
the z-axis is the same as the angular velocity of the cage, ωc .
The mass moment of inertia of a rolling element with respect to
its mass center is neglected.
(4) The movements of the outer ring, inner ring, and rolling
elements are in the radial XY plane and axial z-direction. The
parameters that can describe the position of the bearing compo-
nents are as follows: for the inner ring/arbor, translational DOFs
are uix in the radial x-direction and uiy in the radial y-direction;
for the rolling elements, the movement of each rolling element
in radial XY plane is described in a cylindrical coordinate and
its position is written as 0.5Dm + ubr j (Dm : the pitch diameter),
and the translational DOF of each rolling element in the axial
z-direction is ubzj ; for the outer ring, 3-translational DOFs are
uox , uoy and uoz in radial x-direction, radial y-direction, and axial
z direction respectively.
(5) Because the damping due to friction and a small amount
of lubrication and the structural damping are very small, these
dampings are ignored.
Based on the relative positions of the jth rolling element,
inner race and outer race in [1, 31], the instantaneous curva-
ture center positions of raceways and the jth rolling element
in the case where the rolling elements’, inner ring’s and outer
ring’s displacements are considered are shown in Fig. 3. Con-
sidering bearing components waviness, the instantaneous mu-
tual approaches between the curvature centers of the jth rolling
element and outer-/inner- raceways are respectively as follows
(1) and (2) [1, 32].
δ jib = Dibj − Dib
0 + pi j + ui j ,
Dib
0 = ( f i − 0.5) Db ,
 2  2
Dibj = Dib
rj + Dib
zj ,
(1)
Dib i b
r j = ( f i − 0.5) Db cos α0 − ur j + ur j ,
Dib b
z j = ( f i − 0.5) Db sin α0 + uz j ,
uir j = uix cos (θ j ) + uiy sin (θ j ) ;
5193
 P.P.
P.P. Hou,
Hou, L.Q.
L.Q. Wang,
Wang, and Q.Y. Peng
and Q.Y. Peng

Fig. 3. Positional relationship between the jth rolling element center and raceways curvature center

δ job = Dob ob
j − D0 + po j + uo j The total potential energy includes the potential energy of the
linear springs (Ksx , Ksy ) and the potential energy of the nonlin-
Dob
0 = ( f 1 − 0.5) Db , ear springs (K ib ob
j , K j ). The total potential energy is expressed
   2 as follows (4).
2
Dob
j = Dob
rj + Dob
zj
(2) 1  2 1  2
V = Ksy uiy + Ksx uix +
2 2
Dob b o
r j = ( f o − 0.5) Db cos α0 − ur j + ur j (4)
Nb
2  5/2 Nb 2  5/2
+ ∑ K ibj δj
ib
+ ∑ K ob δ job .
Dob b o
z j = ( f o − 0.5) Db sin α0 − uz j + uz j=1 5 j=1 5
j

uor j = uox cos (θ j ) + uoy sin (θ j ) ,
where Dob 0 is the initial distance between the center of the cur-
vature of the outer raceway and rolling element center. Dob j is
the instantaneous distance between the center of the curvature
of the outer raceway and the jth rolling element center. Dib 0 is
the initial distance between the center of the curvature of the
inner raceway and the rolling element center. Dibj is the instan-
taneous distance between the center of the curvature of the in-
ner raceway and the jth rolling element center. pi j and po j are
the waviness of the inner raceway and outer raceway contacting
with the jth rolling element. ui j and uo j are the waviness of the
jth rolling element contacting with the inner raceway and outer
raceway. Db is the diameter of the rolling elements. fi is the
groove curvature coefficient of the inner race. fo is the groove
curvature coefficient of the outer race. α0 is the initial contact
angle. θ j is the circumferential position of the jth rolling ele-
ment.
The contact force between the jth rolling element and race-
ways can be generated only when there is a compression in the
contact, that is, δ jib > 0 and δ job > 0.
Lagrange’s equation (3) deduces the differential equations of
motion of the bearing components.
  
∂ (∂ T /∂ { ṗ}) ∂t − ∂ T ∂ {p} + ∂V ∂ {p} = { f } , (3)
The total kinetic energy is the sum of the translational and ro-
tational kinetic energy of the arbor, inner ring, rolling elements,
and outer ring. The total kinetic energy is expressed as (5):
T = T r + T i + T b + T o, (5)
where T r is the kinetic energy of the arbor. T i is the kinetic
energy of the inner ring. T b is the kinetic energy of the rolling
elements. T o is the kinetic energy of the outer ring.
The contribution of the arbor contains the translational ki-
netic energy in the radial x and y directions and the rotational
kinetic energy around the z axis. The kinetic energy expression
of the arbor is (6).
1 1  2  2 
T r = I r ωin2 + mr u̇ix + u̇iy , (6)
2 2
where I r is the moment of inertia of the arbor. mr is the mass of
the arbor.
The contribution of the inner ring includes the translational
kinetic energy in the radial x and y directions and the rotational
kinetic energy around the z-axis. The kinetic energy expression
of the inner ring is (7).
1 1  2  2 
T i = I i ωin2 + mi u̇ix + u̇iy , (7)
2 2

where I i is the moment of inertia of the inner ring. mi is the

where T is the kinetic energy; V is the potential energy; {p} is mass of the inner ring.
a vector for generalized degree-of-freedom; { f } is a vector for The contribution of the jth rolling element is the total of the
generalized external forces. translational kinetic energy in the radial direction expressed by

520
4 Bull. Pol.
Bull. Pol.Ac.:
Ac.:Tech.
Tech.68(3)
68(3) 2020
2020
 Vibration
Vibration analysis
analysis of of ball
ball bearingconsidering
bearing considering waviness
wavinessunder
underhigh speed
high andand
speed an axial loadload
an axial

Nb  3/2
a cylindrical coordinate, the translational kinetic energy in the mo üox + ∑ K ib ib
∂ δ jib /∂ uox +
j δj
axial z-direction, and the rotational kinetic energy around the j=1
z-axis. The kinetic energy expression of the jth rolling element Nb  3/2
is (8). + ∑ K ob
j δ job ∂ δ job /∂ uox = 0,
 j=1
1 b  b 2  ḃ 2  2 
T j = m j u̇r j + u̇z j + 0.5Dm + ur j ωc2 , (8)
b b Nb  3/2
2 mo üoy + ∑ K ib
j δj
ib
∂ δ jib /∂ uoy +
j=1
where T jb is the kinetic energy of the jth rolling element. mbj is Nb  3/2 (13)
the mass of the jth rolling element. +∑ K ob
j δ job ∂ δ job /∂ uoy = 0,
Then the kinetic energy of the Nb rolling element is as fol- j=1
low (9): Nb  3/2
Nb
mo üoz + ∑ K ib
j δj
ib
∂ δ jib /∂ uoz +
Tb = ∑ Tjb . (9) j=1
j=1 Nb  3/2
+ ∑ K ob
j δ job ∂ δ job /∂ uoz = Fa ,
The contribution of the outer ring consists of the transla- j=1
tional kinetic energy in the radial x and y directions and ax-
ial z-direction. The kinetic energy expression of the outer ring where e is the eccentricity of the arbor center of mass along its
is (10). geometric center of rotation.
1   2  2 
T o = mo (u̇ox )2 + u̇oy + u̇oz , (10)
2
where mo is the mass of the outer ring. 3. Simulation results
Generalized external forces include the imbalance force due
to the eccentricity of the arbor center of mass along its geomet- This research studies the velocity vibration response character-
ric center of rotation exciting the inner ring/arbor in radial x and istic of the outer ring in the radial x-direction under an axial
y directions and a constant axial load Fa acting on the outer ring load. The main parameters of ball bearing are listed in Table 1.
in the axial z-direction. In the bearing industry, the periodicity of waviness is defined
For the generalized coordinates of the inner ring (uix and uiy ), as a certain number of undulations per revolution (UPR). In the
the governing equations are (11). For the generalized coordi- bearing dynamic analysis, the mathematical modeling of wavi-
nates of the jth rolling element (ubr j and ubzj ), the governing ness is a superimposing of the sinusoidal functions as shown
equations are (12). For the generalized coordinates of the outer in (14) [1].
∞
ring (uox , uoy , and uoz ), the governing equations are (13). W (θ ) = ∑ Al cos (lθ + ϕl ) , (14)
l=1
  Nb  3/2
mi + mr üix + Ksx uix + ∑ K ib
j δj
ib
∂ δ jib /∂ uix + where l is the waviness order. The parameter Al is the magni-
j=1 tude of the lth order. θ is the circumferential position. Expres-
Nb  3/2 sions of θ of pi j , po j , ui j and uoj in (1) and (2) refer to [31]. The
+ ∑ K ob δ job ∂ δ job /∂ uix =
j phase ϕl is uniformly distributed over the interval [0, 2π].
j=1
 
= mi + mr × e × ωi2 cos (ωit) ,
Table 1
(11)
   3/2
Nb Main parameters of ball bearing
mi + mr üiy + Ksy uiy + ∑ K ib
j δj
ib
∂ δ jib /∂ uiy +
j=1 Number of the rolling element, Nb 17
Nb  3/2
Rolling element diameter, Db 3.175 mm
+ ∑ K ob
j δ job ∂ δ job /∂ uiy =
j=1 Density of the rolling element, ρb 7800 kg/m3
 i r

= m +m × e × ωi2 sin (ωit) , Pitch diameter, Dm 26 mm
  Groove curvature coefficient of the outer race, fo 0.535
mbj übr j − mbj × 0.5Dm + ubr j × ωc2 +
 3/2 Groove curvature coefficient of the inner race, fi 0.545
+ K ib
j δj
ib
∂ δ jib /∂ ubr j + Initial contact angle, α0 15◦
 3/2
+ K ob
j δ job ∂ δ job /∂ ubr j = 0, (12) When analyzing the influence of the rotational speed on the
 3/2 bearing vibration response characteristics, the waviness param-
mbj übzj + K ib
j δj
ib
∂ δ jib /∂ ubzj + eters of the bearing component are considered at the same time.
 3/2 The waviness parameters of bearing components have the fol-
+ K obj δ job ∂ δ job /∂ ubzj = 0, lowing characteristics:

Bull. Pol. Ac.:

Bull. Pol. Ac.:Tech.
Tech.68(3)
68(3)2020
2020 5215
 P.P.
P.P. Hou,
Hou, L.Q.
L.Q. Wang,
Wang, and Q.Y. Peng
and Q.Y. Peng

(1) The waviness order is finite. According to JB/T 6642 (the
standard of China), the outer raceway has a maximum order of
250. The inner raceway and rolling element waviness have a
maximum order of 150.
(2) Fig. 4 showed the measurement waviness profile of the
inner raceway of an angle contact rolling bearing and its dis-
crete Fourier transform. The higher the order, the smaller the
amplitude. The relationship between waviness amplitude and
order is satisfied with the following (15) [1].
A cage rotation frequency ( fc ) and the rolling element spinning
Al = , (15)
ls
where the parameter A is the magnitude of the first order. The
exponent s is the amplitude decay for subsequent wave num-
bers.
(3) Considering the waviness of the inner raceway, outer
raceway, and rolling elements simultaneously, the magnitude
of the first order (A) and the their exponent (s) are the same.
The MATLAB ode23tb is used to solve the nonlinear dif-
ferential equations ((11)–(13)). The calculation step size is
1 × 10−6 s. RMS of the amplitude of the outer ring velocity
response will be used to evaluate the bearing vibration level.
The response spectrum of the outer ring velocity response is
obtained by FFT.
3.1. Characteristics of principal frequency caused by wavi-
ness. The relationship between the waviness order and the
principal frequencies are listed in Table 2. Because both the
frequency ( fb ) are proportional to the inner ring/arbor rota-
tion frequency ( fs ), the principal frequencies are a function of
the inner ring/arbor rotational frequency ( fs ) and waviness or-
der (l). Take waviness order 16 as an example. The principal
frequency is satisfied by the function y = 16 × fc − fc as shown
in Fig. 5. Except for the rotational speed of 800 rev/s, the am-
plitude of 16 × fc − fc increases with the increase of fs .

a)
Amplitude (m/s) 40
y=16×fc -fc

20

0
8000
6000 1000
4000
400 600 800
2000 200
Frequency (Hz) Rotational speed (rev/s)

Fig. 5. Influence of rotational speed on the principal frequency in

b)
the radial x-direction (waviness order, amplitude, and phase are 16,
0.5 µm, and 0 rad, respectively)

3.2. Influence of rotational speed on bearing vibration con-

Fig. 4. a) Waviness measurement of the inner raceway of an angle con-
tact rolling bearing. b) Discrete Fourier transform of waviness profile
sidering bearing components waviness. When considering
the bearing components waviness, the parameter A of pi j , po j ,
ui j and uo j is 0.25 µm, the decay exponent s of pi j , po j , ui j and
uo j is 4, and the phase ϕl of pi j , po j , ui j and uo j is 0 rad for each
order. The rotational speed is varied in thirty steps in the range
of 33.3–1000 rev/s. Fig. 6 shows the influence of rotational
speed on spectra and RMS of the outer ring velocity response.

Table 2
Relationship between the waviness order and the principal frequencies (the rotational speed is 400 rev/s; Fa = 60 N; Ksx = Ksy =3e7 N/m;
e = 0 µm; waviness amplitude is 0.5 µm; the waviness phase is 0 rad)

Waviness order (l) n × Nb n × Nb − 1 n × Nb + 1 Direction

n × Nb × fc ± fc n × Nb × fc − fc n × Nb × fc + fc Radial
Due to outer raceway waviness
n × Nb × fc – – Axial
The principal n × Nb × ( fs − fc ) ± fc n × Nb × ( fs − fc ) + fc n × Nb × ( fs − fc ) − fc Radial
Due to inner race waviness
frequencies n × Nb × ( fs − fc ) – – Axial
– – – Radial
Due to rolling element waviness
n × fb – – Axial
where fb is the rolling element spinning frequency.

522
6 Bull. Pol.
Bull. Pol.Ac.:
Ac.:Tech.
Tech.68(3)
68(3) 2020
2020
 Vibration
Vibration analysis
analysis of of ball
ball bearingconsidering
bearing considering waviness
wavinessunder
underhigh speed
high andand
speed an axial loadload
an axial

Frequencies with larger amplitudes, which are represented by 4. Ball bearing high speed vibration test
brighter colors in Fig. 6a), do not exceed 30 kHz. RMS of 0–
30 kHz vs. Rotational speed curve and RMS of the full band vs. Ball bearing vibration test was carried out on rolling bearing
Rotational speed curve overlaps as shown in Fig. 6b). 0–30 kHz high speed vibration measurement rig, whose design guide-
as shown by FB1 (Frequency band) in Fig. 6a) are not affected lines were based on the GB/T 24610 (the standard of China).
by the rotational speed. The process of establishing the test rig was in [33]. The
method of the acceleration response measurement of the bear-
a) #104 ing outer ring was the contact type. This rolling bearing high
0 1
speed vibration measurement rig complied with the princi-
1 300 rev/s 0.8 ple of contact measurement as shown in Fig. 1a). Rotational
Frequency (Hz)

800 rev/s speed was continuously variable by using the frequency trans-
2 FB 1: 0.6 former. The inner ring of tested bearing was driven directly
0-30kHz by the high speed motorized spindle, whose maximum speed
3 0.4
is 1000 rev/s. OptoMET Nova Series Digital Laser Doppler
4 RSR1: RSR2 : RSR3 : 0.2 vibrometer is a noncontact sensor and replaces the contact
33.3-466.7 rev/s 466.7-633.3 rev/s 633.3-1000 rev/s sensor for measuring the outer ring vibration. This vibrome-
5 0
0 200 400 600 800 1000 ter can output displacement, velocity, and acceleration signals.
Rotational speed (rev/s) Rolling bearing high speed vibration measurement rig with Op-
toMET Nova Series Digital Laser Doppler vibrometer con-
b) 0.04 formed to the principle of bearing vibration measurement in
RSR1 : 33.3-466.7 RSR3 : 633.3-1000 Fig. 1b). The image of the noncontact and contact measure-
0.03 rev/s Full band rev/s ment of the rolling bearing high speed vibration measurement
RMS (m/s)

0-30 kHz rig was shown in Fig. 7. The test bearing was high speed an-
0.02 gle contact rolling bearings (7003 series made by FAG). Test
300 rev/s
533.3 speeds were 25/60/100/200/300/400/500/600 rev/s and an ax-
0.01 rev/s
RSR2 : 466.7-633.3 800 rev/s ial load was 60 N. DT9847 was used for data acquisition and
0 rev/s
the sampling frequency was 200 kHz. Lubricant oil was MO-
0 200 400 600 800 1000 TOREX SPINDLE LUBE 68.
Rotational speed (rev/s)

Fig. 6. Simulated results in the radial x-direction at Ksx = Ksy =

3e11 N/m and e = 0 µm: a) Influence of rotational speed on veloc-
ity spectra of the outer ring. b) Influence of rotational speed on RMS
of the outer ring velocity response

However, in FB1 , the distribution of frequencies with larger

amplitudes changes with increasing of rotational speed. The
rotational speed at which the frequency distribution changes
in Fig. 6a) is consistent with the rotational speed at which
the RMS is abrupt in Fig. 6b), namely 300 rev/s, 466.7 rev/s,
633.3 rev/s, and 800 rev/s. When the rotational speed range
(RSR) is 33.3–466.7 rev/s (RSR1 in Fig. 6a)), the distribu-
tion of frequencies with larger amplitudes is within 10 kHz–
20 kHz and RMS in RSR1 rises slowly with rotational speed
increasing except for the rotational speed of 300 rev/s. The
frequency bandwidth with larger amplitudes at 300 rev/s is
a little bit wider and RMS of 300 rev/s is outstanding. In
RSR2 , frequencies of larger amplitudes are dispersed in FB1 Fig. 7. Noncontact and contact measurement of the rolling bearing
and RMS increases from 466.7 rev/s to 533.3 rev/s but falls high speed vibration measurement rig
in 533.3–633.3 rev/s. In RSR3 , except for the rotational speed
of 800 rev/s, the distribution of frequencies with larger ampli-
tudes is bifurcated. RMS in RSR3 declines slowly with rota- The test bearing is connected to the drive spindle and the
tional speed increasing but the rotational speed of 800 rev/s. loading tool. In addition to the bearing itself, the vibration of
This is because the condition of frequencies with larger bearing outer ring is mainly superimposed on the vibration of
amplitudes at 800 rev/s is similar to that of RSR2 . There- the contact part of the bearing and the vibration generated by the
fore, RMS of 800 rev/s is larger than RMS of other speed interaction between parts of the system, such as the loading tool
in RSR3 . in contact with the outer ring, the inner ring mounted on drive

Bull. Pol. Ac.:

Bull. Pol. Ac.:Tech.
Tech.68(3)
68(3)2020
2020 5237
 P.P. Hou, L.Q. Wang, and Q.Y. Peng
P.P. Hou, L.Q. Wang, and Q.Y. Peng

spindle’s rotor, which is also supported by multiple bearings, and the outer ring is deflected, when an axial force is applied.
as well as the influence of misalignment between the externally These are the same as variable compliance vibration [34, 35].
applied load and the spindle axis of rotation. Fig. 8 shows the This phenomenon can be confirmed by theoretical calculation.
frequency-domain comparison of contact and noncontact mea- A 10 N radial load in the x-direction of the outer ring is applied.
surements at 400 rev/s. It can be seen from Fig. 8 that high mul- When the order of the inner and outer raceway waviness is Nb
tiple harmonic frequencies of the rotational speed are prominent respectively, the simulated results show in Fig. 10. There are
in the response spectrum of the contact measurement, but not Nb × fc , 2 × Nb × fc in Fig. 10a) and Nb × ( fs − fc ) in Fig. 10b).
in the response spectrum of the noncontact measurement. This These show that the simplification of the bearing model is rea-
means that for the contact measurement, at high speed the influ- sonable. On the other hand, the rolling bearing high speed vi-
ence of the system components on the vibration of the bearing bration measurement rig can be used to measure the bearing
itself is much more obvious than those of the noncontact mea- vibration characteristics caused by waviness, which indicates
surement. Therefore, from the perspective of analyzing the vi- that the bearing outer ring vibration response obtained by the
bration characteristics of the bearing itself, the noncontact mea- experiment reflects the vibration characteristics of the individ-
surement is far more suitable than the contact measurement to ual bearings to some extent.
evaluate the bearing vibration level at high speed.
a) 6
10
contact measurement N b ×f c
Amplitude (m/s 2)

Amplitude (mm/s)
noncontact measurement 4 2N b ×f c
fc
5 30f s N b ×f c +f c
N b ×f c -f c
10f s 40f s 2
fs 20f s 50f s
0 0
0 1 2 3 4 5 0 2000 4000 6000 8000 10000
Frequency (Hz) #104 Frequency (Hz)
Fig. 8. Comparison of outer ring response spectra of contact and non- b) 6
contact measurements at 400 rev/s

4 f
Amplitude (mm/s)

4.1. Validation of ball bearing model. At 400 rev/s, the ve- c

locity spectrum of the test bearing is shown in Fig. 9. Run- N b ×(f s -f c )

N b ×(f s -f c ) N b ×(f s -f c )
ning speed is a little different from theoretical speed. There is a 2
clear rotational frequency ( fs ≈ 399 Hz) and a double frequency -f c +f c

(2 fs ≈ 798.8 Hz). Harmonics of rotation frequency (3 fs , 4 fs ,

5 fs , 6 fs ) also exist. According to Table 2, there are princi- 0
0 2000 4000 6000 8000 10000
pal frequencies, such as Nb × fc + fc , Nb × ( fs − fc ) − fc and
Frequency (Hz)
Nb × ( fs − fc ) + fc .
However, in Fig. 9, there are Nb × fc and Nb ×( fs − fc ), which Fig. 10. Simulated velocity response spectra of the outer ring due to the
are not in Table 2. This is because the outer ring is not tilted un- inner and outer raceway waviness of order l = Nb when the outer ring
der an axial force in the theoretical calculation. However, in the is subjected to 10 N x-radial force and 60 N z-axial force at 400 rev/s:
experiment, the axial force is not aligned with the arbor axis a) Waviness in the outer raceway. b) Waviness in the inner raceway

8
f s =399Hz Vertical coordinate enlargement
0.5
N b ×(f s -f c )+f c
0.4 N b ×(f s -f c )
6
Amplitude (mm/s)

N b ×f c +f c 3392Hz
0.3

4 2181Hz N b ×f c
0.2 N b ×(f s -f c )
fc
4f s 6f s -f c
0.1 176.2Hz 3f s 5f s
2 0
0 500 1000 1500 2000 2500 3000 3500 4000
2f s

0
0 500 1000 1500 2000 2500 3000 3500 4000
Frequency (Hz)

Fig. 9. Outer ring velocity response spectrum of at 400 rev/s on the rolling bearing high speed vibration measurement rig

524
8 Bull. Pol.
Bull. Pol.Ac.:
Ac.:Tech.
Tech.68(3)
68(3) 2020
2020
 Vibration
Vibration analysis
analysis of of ball
ball bearingconsidering
bearing considering waviness
wavinessunder
underhigh speed
high andand
speed an axial loadload
an axial

4.2. Experimental results. As shown in Fig. 9, fs and 2 fs are
prominent. To eliminate the effects of arbor eccentricity and
misalignment external load, the lowest cutoff frequency is 2 fs .
Fig. 11 shows the experimental results of spectra and RMS of
the outer ring velocity response. Compared with Fig. 6, exper-
imental and simulated results exhibit similar and different be-
haviors. Below 300 rev/s, the minimum frequency band, whose
RMS is almost equal to RMS of the full band, is 0–30 kHz, as
shown by FB2 in Fig. 11a). After the rotational speed exceeds
300 rev/s, the difference value between RMS of 0–30 kHz and
RMS of the full band gets larger. In FB2 , frequencies with larger
amplitudes are mainly distributed around 15 kHz and around
25 kHz, which is different from the case where the distribution
characteristic of the frequencies with larger amplitudes in FB1
of Fig. 6a) change as the rotational speed increases. These fre-
quencies with larger amplitudes, which are not affected by the
rotational speed either, have been confirmed to be the natural
frequencies of the bearing system [15, 16]. On the other hand,
the tendency of RMS – rotational speed curve in Fig. 11b) is
also similar to that of Fig. 6b). The differences are that rota-
tional speed corresponding to maximum RMS in Fig. 11b) is
300 rev/s, while in Fig. 6b) this rotational speed is 533.3 rev/s,
and that the value of RMS of the outer ring velocity response
of the experimental results is significantly different from the
RMS of the outer ring velocity response of the simulated re-
sults. Because the bearing model has simplified many of the
effects of the test rig, compared to the experiment, the theoreti-
cally calculated outer ring vibration response is more reflective
a) 44
0 #10
0.04
(Hz)
Frequency (Hz)
of the structural vibration of the individual bearing itself. When
the bearing becomes part of a larger assembly, the measured
bearing outer ring vibration response contains various vibra-
tion response originated each one of the parts of the system.
The vibration of rotating rolling bearings is a complex physical
phenomenon dependent on the conditions of operation, and a
change in the rotational speed alone causes a strong change in
the excitation of the dynamics of all parts of the system [17–19,
36]. These lead to significant differences between theoretical
and experimental results.
Simulated and experimental results are of great importance
to the high speed vibration evaluation of rolling bearings.
Firstly, it is well known that one of the primary goals of bear-
ing vibration measurement is the ability to evaluate the char-
acteristics of waviness. In ISO 15242, at the default mea-
surement speed (30 rev/s), the total analytical frequency (TF)
band is from 50 Hz to 10 000 Hz, and it is also divided into
three frequency bands: low frequency (LF) band is 50–300 Hz,
medium frequency (MF) band is 300-1800 Hz, and high fre-
quency (HF) band is 1800–10 000 Hz. Principal frequencies
caused by waviness are proportional to rotational speed. There-
fore, when tested speed is increased, all frequency bands have
to be expanded in the same proportion theoretically to in-
clude the frequency characteristics caused by the waviness.
Because these periodic frequencies are proportional to rota-
tional speed, based on the characteristics of node frequen-
cies of default analysis frequency band of 30 rev/s, the cut-
off frequencies of these bands could be converted into param-
eters depended on the rotational speed, namely 2 fs − 300 fs
(TF band), 2 fs − 10 fs (LF band), 10 fs − 60 fs (MF band) and
30 fs − 300 fs (HF band). Fig. 12 shows the experimental veloc-
ity response spectrum of 0–300 Hz at 25 rev/s. It also demon-

strates that vibration characteristics caused by waviness can

FB 22 : 0-30 kHz
Frequency

5 be reflected in 0–300 Hz. Therefore, the division of the fre-

0.02 quency band will depend on the test speed. For the tested
bearing, at high speed and low speed, 2 fs − 10 fs is help-
300 rev/s
10 0 ful to analyze the frequency characteristics caused by the
30 60 100 200 300 400 500 600 waviness order which is close to the number of rolling ele-
Rotational speed (rev/s) ments.
While the maximum cutoff frequency of TF at high speed
will get great larger than that of 30 rev/s, for example, at
b) 1.5 400 rev/s, the maximum cut-off frequency is 120 kHz, and the
2f s -100kHz
s sampling frequency will be more than 240 kHz, which will be
2f s -30kHz
s beyond the capability of most acquisition systems. The results
(mm/s)
RMS (mm/s)

1 of Fig. 6 and Fig. 11 show that the frequency bandwidth close

300 to the vibration energy of the full band is limited. The cut-off
rev/s frequency of FB1 and FB2 is much lower than 120 kHz. RMS
RMS

0.5 of the frequency band is equal to that of the full band and re-
flects the influence of rotational speed on the bearing vibration
level. On the other hand, according to [1] and the characteris-
0
tics of the measured waviness in Fig. 4, the higher the order
0 100 200 300 400 500 600
of waviness, the smaller its amplitude. The smaller the wavi-
Rotational speed (rev/s)
ness amplitude is, the lesser the impact on the amplitude of the
Fig. 11. Experimental results in radial x-direction: a) influence of rota- bearing vibration response [13]. Therefore, it is not necessary
tional speed on spectra distribution of the outer ring velocity response to satisfy the theoretical maximum cutoff frequency of TF. For
(frequency band: 2 fs − 100 kHz), b) influence of rotational speed on example, for the tested bearing, the maximum cutoff frequency
RMS of the outer ring velocity response of TF could be 30 kHz at high speed.

Bull. Pol. Ac.:

Bull. Pol. Ac.:Tech.
Tech.68(3)
68(3)2020
2020 5259
 P.P.
P.P. Hou,
Hou, L.Q.
L.Q. Wang,
Wang, and Q.Y. Peng
and Q.Y. Peng

0.5
f s =23.91Hz #10-3 Vertical coordinate enlargement
5
N b ×f c +f c N b ×(f s -f c )
0.4
4 or 2f b or (Nb + 1)×(f s -f c )-f c
(Nb + 1)×f c
Amplitude (mm/s)

N b ×f c +f c
0.3 3 fb
N b ×f c -f c 3f b

2 (Nb -1)×f c -f c N b ×(f s -f c ) N b ×(f s -f c )

0.2 -f c +f c
1

0.1 0
0 2f s 50 100 150 200 250 300

0
0 50 100 150 200 250 300
Frequency (Hz)

Fig. 12. Outer ring velocity response spectrum at 25 rev/s on BVT-5 (standard rolling bearing vibration measurement instrument made
by HANGZHOU BEARING TEST & RESEARCH CENTER CO., LTD.)

5. Conclusions [2] S. Adamczak, R. Domagalski, E. Sender, P. Zmarzły, and

According to the principle of bearing vibration measurement,
a 2Nb + 5DOF dynamic nonlinear model of the ball bearing is
proposed. The mathematical bearing model involves the mo-
tions of the inner ring in the radial XY plane, the outer ring and
rolling elements in the radial XY plane and axial Z direction.
Bearing components waviness and rotational speed are consid-
ered into the ball bearing model. In the case where the inner
ring raceway, outer ring raceway, and rolling elements wavi-
ness are considered at the same time, characteristics of response
spectra and RMS of ball bearing vibration with the increase
of the rotational speed are discussed. The vibration experi-
ments of a high speed bearing are performed on the ball bear-
ing high speed vibration measurement rig under low, medium
and high speed. After eliminating the effects of the rotational
frequency and its double frequency, simulated and experimen-
tal results showed that the width of the minimum frequency
band whose RMS is almost equal to RMS of the full band is
not influenced by the rotational speed, while RMS of the full
band will fluctuate with the increase in rotational speed. This
width of the minimum frequency band affects the choice of
the sensor’s performance parameters and vibration measure-
ment mode and determines the analysis frequency bandwidth
of the evaluation method of bearing high speed vibration mea-
[10] C.K. Babu, N. Tandon, and R.K. Pandey, “Vibration model-
surement, which has a great significance for engineering appli-
cations.
Acknowledgements. The investigations were supported by
[11] Y. Zhuo, X. Zhou, and C. Yang, “Dynamic analysis of double-
the National Natural Science Foundation of China (Research
Project U1637206) and the National key research and develop-
ment program (Research Project 2018YFB0703804).
[12] S.P. Harsha and C. Nataraj, “The effect of surface waviness and
R EFERENCES
[1] J.A. Wensing, “On the dynamics of ball bearings”, pp. 99–101,
47–97, 159–162, 16–18, Department of Mechanical Engineer-
ing, University of Twente, 1998.
Ł. Gorycki, “Research methods and testing stand developed to
examine vibrations generated by rolling bearing”, Diagnostyka.
17 (1), 41–49 (2016).
[3] E. Yhland, “A linear theory of vibrations caused by ball bearings
with form errors operating at moderate speed”, ASME. J. Tribol.
114 (2), 348–359 (1992).
[4] K. Ono and K. Takahasi, “Theoretical analysis of shaft vibra-
tion supported by a ball bearing with small sinusoidal waviness”,
IEEE Trans. Magn. 32 (3), 1709–1714 (1996).
[5] Nizami Aktürk, “The effect of waviness on vibrations associated
with ball bearings”, J. Tribol. 121 (4), 667–677 (1999).
[6] N. Tandon and A. Choudhury, “A theoretical model to predict the
vibration response of rolling bearings in a rotor bearing system
to distributed defects under radial load”, J. Tribol. 122 (3), 609–
615 (2000).
[7] G.H. Jang and S.W. Jeong, “Nonlinear excitation model of ball
bearing waviness in a rigid rotor supported by two or more ball
bearings considering five degrees of freedom”, J. Tribol. 124 (1),
82–90 (2002).
[8] G. Jang and S-W. Jeong, “Vibration analysis of a rotating system
due to the effect of ball bearing waviness”, J. Sound Vibr. 269 (3-
5), 709–726 (2004).
[9] G.H. Jang and S.W. Jeong, “Analysis of a ball bearing with wavi-
ness considering the centrifugal force and gyroscopic moment of
the ball”, J. Tribol. 125 (3), 487–498 (2003)
ing of a rigid rotor supported on the lubricated angular contact
ball bearings considering six degrees of freedom and waviness
on balls and races”, J. Vib. Acoust. 134 (1), 011006-011006-12
(2012).
row self-aligning ball bearings due to applied loads, internal
clearance, surface waviness and number of balls”, J. Sound Vibr.
333 (23), 6170–6189 (2014).
number of rolling elements on the dynamic behavior of a rotor-
bearing system”, Proceedings of the ASME 2007 International
Design Engineering Technical Conferences and Computers and
Information in Engineering Conference. Volume 1: 21st Biennial
Conference on Mechanical Vibration and Noise, Parts A, B, and
C. Las Vegas, Nevada, USA, 2007, pp. 1755–1762.

526
10 Bull. Pol.
Bull. Pol.Ac.:
Ac.:Tech.
Tech.68(3)
68(3) 2020
2020
 Vibration
Vibration analysis
analysis of of ball
ball bearingconsidering
bearing considering waviness
wavinessunder
[13] Y. Shao, P. Wang, and Z. Chen, “Effect of waviness on vibration
and acoustic features of rolling element bearing”, ASME 2012
International Design Engineering Technical Conferences and
Computers and Information in Engineering Conference. Amer-
ican Society of Mechanical Engineers, Chicago, Illinois, USA,
2012, pp. 165–170.
[14] S. Adamczak and P. Zmarzły, “Influence of raceway waviness
on the level of vibration in rolling-element bearings”, Bull. Pol.
Ac.: Tech. 65 (4), 541–551 (2017).
[15] H. Ohta and S. Satake, “Vibrations of hybrid ceramic ball bear-
ings”, J. Sound Vibr. 192 (2), 481–493 (1996).
[16] H. Ohta and S. Satake, “Vibrations of the all-ceramic ball bear-
ing”, J. Tribol. 124 (3), 448–460 (2002).
[17] H. Cao, F. Shi, Y. Li, B. Li, and X. Chen, “Vibration and stability
analysis of rotor-bearing-pedestal system due to clearance fit”,
Mech. Syst. Signal Proc. 133, 106275 (2019).
[18] S.P. Harsha, “Nonlinear dynamic analysis of a high-speed rotor
supported by rolling element bearings”, J. Sound Vibr. 290 (1-2),
65–100 (2006).
[19] A. Sharma, N. Upadhyay, P. Kumar Kankar, and M. Amarnath,
“Nonlinear dynamic investigations on rolling element bearings:
A review”, Adv. Mech. Eng. 10 (3), 1–15 (2018).
[20] S. Khanam, N. Tandon, and J.K. Dutt, “Multi-event excitation
force model for inner race defect in a rolling element bearing”,
ASME. J. Tribol. 138 (1), 011106 (2016).
[21] S. Sassi, B. Badri, and M. Thomas, “A numerical model to pre-
dict damaged bearing vibrations”, J. Vib. Control. 13 (11), 1603–
1628 (2007).
[22] N.S. Feng, E.J. Hahn, and R.B. Randall, “Using transient analy-
sis software to simulate vibration signals due to rolling element
bearing defects”, Appl. Mech: Progress and Applications. 689–
694 (2002).
[23] D. Petersen and C. Howard, “Bearing defect size estimation
for extended raceway defects”, INTER-NOISE and NOISE-CON
Congress and Conference Proceedings. Institute of Noise Con-
trol Engineering. 249 (5), 2787–2796 (2014).
[24] N. Sawalhi and R.B. Randall, “Simulating gear and bearing in-
teractions in the presence of faults: Part I. The combined gear
bearing dynamic model and the simulation of localised bearing
faults”, Mech. Syst. Signal Proc. 22 (8), 1924–1951 (2008).
Bull. Pol. Ac.:
Bull. Pol. Ac.:Tech.
Tech.68(3)
68(3)2020
2020
underhigh speed
high andand
speed an axial loadload
an axial
[25] B. Dolenc, P. Boškoski, and Ð. Juričić, “Distributed bearing fault
diagnosis based on vibration analysis”, Mech. Syst. Signal Proc.
66–67, 521–532 (2016).
[26] D. Petersen, C. Howard, N. Sawalhi, A. Moazen Ahmadi, and
S. Singh, “Analysis of bearing stiffness variations, contact forces
and vibrations in radially loaded double row rolling element
bearings with raceway defects”, Mech. Syst. Signal Proc. 50–51,
139–160 (2015).
[27] J. Liu and Y. Shao, “Dynamic modeling for rigid rotor bearing
systems with a localized defect considering additional deforma-
tions at the sharp edges”, J. Sound Vibr. 398 (23), 84–102 (2017).
[28] P. Patra, V. Huzur Saran, and S.P. Harsha, “Non-linear dynamic
response analysis of cylindrical roller bearings due to rotational
speed”, Proceedings of the Institution of Mechanical Engineers,
Part K: J. of Multi-body Dynamics. 233 (2), 379–390 (2018).
[29] M. Tadina and M. Boltežar, “Improved model of a ball bearing
for the simulation of vibration signals due to faults during run-
up”, J. Sound Vibr. 330 (17), 4287–4301 (2011).
[30] L. Cui, Z. Jin, J. Huang, and H. Wang, “Fault Severity Classifi-
cation and Size Estimation for Ball Bearings Based on Vibration
Mechanism”, IEEE Access. 7, 56107–56116 (2019).
[31] X. Li, K. Yu, H. Ma, L. Cao, Z. Luo, and H. Li, et al., “Analysis
of varying contact angles and load distributions in defective an-
gular contact ball bearing”, Eng. Fail. Anal. 91, 449–464 (2018).
[32] D.S. Shah and V.N. Patel, “Theoretical and experimental vibra-
tion studies of lubricated deep groove ball bearings having sur-
face waviness on its races”, Measurement. 129, 405–423 (2018).
[33] B. Yingcun, “Study on vibration characteristics of grease-
lubricated bearings used in high-speed spindle”, Harbin Institute
of Technology. (2015) [in Chinese].
[34] R. Yang, L. Hou, Y. Jin, Y. Chen, and Z. Zhang, “The Varying
Compliance Resonance in a Ball Bearing Rotor System Affected
by Different Ball Numbers and Rotor Eccentricities”, J. Tribol.
140 (5), 051101 (2018).
[35] H. Cheng, Y. Zhang, W. Lu, and Z. Yang, “Research on time-
varying stiffness of bearing based on local defect and varying
compliance coupling”, Measurement. 143, 155–179 (2019).
[36] R. Serrato, M.M. Maru, and L.R. Padovese, “Effect of lubricant
viscosity grade on mechanical vibration of roller bearings”,
Tribol. Int. 40 (8), 1270–1275 (2007).
527
11